Mechanisms of species coexistence: a field test of theoretical models

Mechanisms of species coexistence: a field test of theoretical models
OIKOS 105: 512 /524, 2004
Mechanisms of species coexistence: a field test of theoretical models
using intertidal snails
Rémy Rochette and Tamara C. Grand
Rochette, R. and Grand, T. C. 2004. Mechanisms of species coexistence: a field test of
theoretical models using intertidal snails. / Oikos 105: 512 /524.
Competitor coexistence is often facilitated by spatial segregation. Traditionally, spatial
segregation is predicted to occur when species differ in the habitat in which they are
either superior at competing for resources or less susceptible to predation. However,
predictions from a behavioural model demonstrate that spatial segregation and
coexistence can also occur in the absence of such interspecific trade-offs in
competitive ability and vulnerability to predation. Unlike other models of competitor
coexistence this model predicts that when species rank both habitat productivity and
‘riskinesses’ similarly, but differ slightly in their habitat-specific vulnerabilities to
predators, they will tend to segregate across habitats, with the species experiencing the
higher ratio of mortality risk across the habitats occurring primarily in the safer
habitat. Here, we investigate the hypothesis that intraspecific trade-offs between
resource availability and mortality risk can lead to spatial segregation of competing
species by (1) documenting the spatial (i.e. intertidal) distribution of two marine snails,
Littorina sitkana and L. subrotundata and (2) performing field experiments to quantify
growth and mortality rates of each species at ‘low’ and ‘high’ intertidal heights. Our
results indicate that both species agree on the rankings of habitat riskiness and
productivity, experiencing higher predation and higher growth in low- than in highintertidal habitats. However, L. sitkana and L. subrotundata experienced differences in
their habitat-specific mortality risks and growth rates. Despite both species being
similarly at risk of predation in high-intertidal habitats (where mortality was lower),
L. subrotundata was subject to significantly higher mortality than L. sitkana at the lowintertidal height. In contrast, growth rate differences between habitats were greater for
L. sitkana than for L. subrotundata . Whereas both species grew at the same rate at the
high-intertidal level (where growth was lower), L. sitkana individuals grew more
rapidly than L. subrotundata snails at the low-intertidal level. As predicted by the
behavioural model, the species that experienced the higher ratio of mortality across
habitats (i.e. L. subrotundata ) occurred exclusively in the safer, high-intertidal habitat.
Taken together, these results provide support for the hypothesis that spatial
segregation, and potentially competitor coexistence, can occur in the absence of
interspecific trade-offs in resource acquisition ability or vulnerability to predation.
Rémy Rochette, Simon Fraser Univ. and Bamfield Marine Sciences Centre, British
Columbia, Canada. Present address: Dept. of Biology, Univ. of New Brunswick, P.O. Box
5050, Saint John, NB. E2L 4L5, Canada (rochette@unbsj.ca). / T. C. Grand, 108 Roe
Drive, Port Moody, British Columbia, Canada, V3H 3M8.
When two species compete for access to common
resources and are at risk of being consumed by shared
predators, their continued coexistence is frequently
facilitated by spatial segregation. Typically, spatial
segregation is predicted to occur when (1) species differ
in the habitat in which they are superior at competing for
resources (MacArthur and Levins 1967, Lawlor and
Maynard Smith 1976, Vincent et al. 1996), (2) species
Accepted 13 October 2003
Copyright # OIKOS 2004
ISSN 0030-1299
512
OIKOS 105:3 (2004)
differ in the habitat in which they are better at avoiding
predators (Kotler 1984, Longland and Price 1991,
Brown 1998), (3) species differ in their perceptions of
safe and risky habitats and each species has a higher
foraging efficiency in its riskier habitat (Brown 1998) and
(4) one species is better at avoiding predators in all
habitats while the other is the superior competitor in all
habitats (Brown 1998). Most such models imply (or
explicitly require) the presence of strong interspecific
trade-offs in traits related to competitive ability and/or
vulnerability to predation. Such trade-offs might be
expected to evolve when the foraging and/or antipredator strategies required for survival in one habitat
differ from those required in other habitats, perhaps
because habitats provide different types of resources or
harbor different species of predators (e.g. benthic and
limnetic stickleback fish, Gasterosteus aculeatus L.;
Vamosi 2002).
Recently, Grand and Dill (1999) and Grand (2002)
suggested that spatial segregation of competing species
can also occur in the absence of such interspecific tradeoffs, and in particular, when the same resources and
predators are present in alternate habitats. This suggestion stems from their theoretical investigations into the
effects of differences in competitive ability and habitatspecific vulnerability to predation on the outcome of
intraspecific foraging /predation risk trade-offs during
habitat selection. Unlike most other models of competitor coexistence (references above), their model predicts
that segregation can occur even when species rank
similarly both the productivity and ‘riskinesses’ of
different habitats; if species differ slightly in their
habitat-specific vulnerabilities to predators, they will
tend to segregate across habitats, with the species
experiencing the higher ratio of mortality risk across
the habitats occurring primarily in the safer habitat.
Thus, the species found at the riskiest location need not
be the one that experiences the lowest risk of mortality
there. This pattern of spatial segregation occurs because
the relative mortality cost of inhabiting the riskier
habitat is offset by increased foraging gains for one
species (the species with the lower ‘risk ratio’) but not the
other, even in the face of resource competition.
Such spatial segregation has been observed in a twospecies system of marine snails (Littorina subrotundata
[Carpenter] and L. sitkana Philippi) that coexist along
the rocky intertidal shores of the northeastern Pacific
Ocean. Whereas L. subrotundata seems to thrive on
wave-exposed headlands and in salt marshes, L. sitkana
predominates on moderately-exposed to sheltered shores
(Reid 1996). Segregation is not absolute, however, and
the two species do co-exist on many shores. There is no
published data on the vertical distribution of these two
species where they coexist on the same shore, but in
sheltered inlets near the Bamfield Marine Sciences
Centre, Barkley Sound (West coast of Vancouver
OIKOS 105:3 (2004)
Island), L. subrotundata seems to predominate in the
high-intertidal and L. sitkana in the mid intertidal (R.
Rochette, pers. obs.). On these shores, snails are subject
to strong predation pressure by crabs (e.g. Cancer
productus Randall), and, to a lesser extent, fish (e.g.
perch; McCormack 1982, Robles et al. 1989, Behrens
Yamada and Boulding 1996, Boulding et al. 1999).
On wave-sheltered shores, littorinid snails face greater
predation risk in the lower, than upper, parts of the
intertidal zone (Behrens Yamada and Boulding 1996,
Rochette and Dill 2000), presumably due to differences
in submergence time related to tidal fluctuations, but
perhaps also because fish and crabs are themselves at
risk of being preyed upon by birds (e.g. herons) and
terrestrial mammals (e.g. otters, minks) if they venture
too close to the surface. Importantly, the greater safety
of higher-intertidal areas seems to come at a cost of
reduced feeding opportunities. Littorina subrotundata
and L. sitkana feed mainly on diatoms and unicellular
algae that they scrape off algae and hard substrates (e.g.
rocks, oyster shells) with their radula, and to a lesser
extent on macroalgae and lichens. To avoid desiccation,
they graze more actively when submerged than emerged,
and reciprocal transplant experiments in Barkley Sound
have shown that L . sitkana grows more rapidly in lower,
than in upper, parts of their intertidal range (McCormack 1982, Rochette et al. 2003).
Here, we investigate the hypothesis that intraspecific
trade-offs between resource availability and mortality
risk can lead to spatial segregation of competing species
by (1) documenting the intertidal distribution of L.
sitkana and L. subrotundata at two study sites and (2)
performing field experiments to quantify growth and
mortality rates of each species at ‘low’ and ‘high’
intertidal heights. Growth and mortality rate data were
used to differentiate between five potential mechanisms
of spatial segregation (Table 1 for a summary of each
hypothesis’ key assumptions and predictions).
Methods
We conducted our study between 6 April 2000 and 16
March 2001, on two gently-sloping gravel beaches in
wave-sheltered Bamfield Inlet (48850?, 125808?), Barkley
Sound (Canada), northeastern Pacific (map in Boulding
et al. 1999). These two shores are separated by approximately 800 m and were chosen because they harbor high
densities of littorinids across a relatively wide range of
intertidal heights (ca 1 to 3 m). The substrate of beaches
up Bamfield Inlet tends to vary with intertidal height,
gradually changing from fine sediments at the lower level
to gravel and bedrock higher in the intertidal. Different
species of vascular plants and macroalgae are normally
present at different levels; the subtidal eelgrass Zostera
marina L. extends up to about the 0-m mark, the
513
514
Table 1. A summary of five spatial segregation mechanisms and their key assumptions and predictions.
Mechanism
Differential competitive
superiority (MacArthur
and Levins 1967)
Differential vulnerability
(Kotler 1984, Brown
1998)
Differential foraging efficiency and vulnerability
to predation (Brown
1998)
Interspecific trade-off in
foraging efficiency and
vulnerability (Brown
1998)
Differential ratios of mortality risk across habitats (Grand and Dill
1999)
Habitat attributes
Species’ perceptions of habitats
Predicted distribution
Resource availability
Risk of mortality
Habitats differ in resource
availability only
Species rank habitats differently
Not considered
Each species in habitat in which it’s
competitively superior
Habitats differ in risk of
mortality only
Not considered (Kotler 1984) or species rank habitats identically (Brown
1998)
Species rank habitats differently; each
species has a higher foraging efficiency
in a different habitat
Species rank habitats differently
Each species in habitat in which it’s least
vulnerable
Habitats differ in risk of
mortality only
Habitats differ in risk of
mortality only
Habitats differ in resource
availability and risk of mortality; high risk habitat most
profitable
Species rank habitats differently; each
is more vulnerable to predation in the
habitat in which its foraging efficiency
is greatest
Species rank habitats identically, how- Species rank habitats identically,
ever, one species is the more efficient however, one species is less vulnerable
forager in both habitats
to predation in both habitats
Species rank habitats identically; species may differ in competitive ability,
but resource acquisition abilities similar in both habitats
Species rank habitats identically;
small, habitat-specific differences in
vulnerability to predators between
species
Both species found in both habitats; each
uses one habitat for food and the other for
safety
One species is found exclusively in a single
habitat, the other occurs in both habitats
Species with the lower ratio of mortality
risk across habitats in the riskier habitat,
regardless of which species is absolutely at
greater risk there
OIKOS 105:3 (2004)
filamentous green algae Enteromorpha intestinalis (L.) is
common between 0 and 1 m, and the brown algae Fucus
distichus L. is common to dominant on boulders and
bedrock between the 2- and 3-m intertidal marks. The
density of littorinids on these shores is partly related to
the abundance of hard substrate, such as cobbles,
boulders and bedrock; littorinids are uncommon on
mud and sand bottoms.
There are four species of littorinids in Bamfield Inlet,
two that undergo benthic larval development inside
gelatinous egg masses (L. sitkana and L. subrotundata )
and two that possess a pelagic larval stage (L. scutulata
Gould and L. plena Gould). In this study, we decided to
focus on the former two species, because we expected
benthic larval development to be more conducive to
consistent (both temporally and spatially) vertical distribution patterns and habitat segregation among species. Furthermore, selecting these two species meant not
having to consider the role of larval dispersal and
settlement in mediating snail distribution. L. sitkana
and L. subrotundata are very similar morphologically,
but they show some differentiation with respect to body
pigmentation (i.e. head, foot, and tentacles), shell
sculpturing, and the shape of female reproductive organs
(Reid 1996). In this study, we used shell sculpturing for
species identification, because it is non-destructive and
more reliable than body pigmentation (R. Rochette, pers.
obs.). The shell of L. sitkana has relatively high ‘‘ridges’’,
or bumps, and fine striations between the ridges, whereas
that of L. subrotundata has relatively long and flat ridges
with no microsculpturing between (Reid 1996). In order
to confirm the accuracy of our identifications based on
shell sculpturing, we dissected more than 40 adult
females of each species and identified them based on
the shape of the palleal oviduct (Reid 1996). Shell
sculpturing proved to be highly reliable as a means of
discriminating the two species at our study sites; the only
two ‘‘mistakes’’ committed were for very large and
eroded shells.
Snail spatial distribution
We determined the vertical distribution of L. sitkana and
L. subrotundata at our two study sites in late April (site
A) and early July (site B) 2000. Each site was 6 m long,
and ranged from 1.2 to 2.8 m above 0 datum (Canadian
Hydrographic Service); approximately 6 and 11 m
separated the 1.2 and 2.8 m levels at sites A and B,
respectively. For ease of comparison with earlier work
done in the same area, intertidal heights were estimated
using as a reference point the site described in Rochette
and Dill (2000); there was a /0.12 m tidal anomaly the
day the reference site was established, but we present
non-corrected (e.g. 2.5 m instead of 2.62 m) tidal
amplitudes throughout the paper for simplicity.
OIKOS 105:3 (2004)
We used random quadrat (0.1/0.1 m) sampling to
evaluate the density of snails at 0.4-m increments in
intertidal height (i.e. 1.2, 1.6, 2.0, 2.4, and 2.8 m) at both
study sites. At each level, we first laid down a measuring
tape parallel to the shoreline, and then used a table of
random numbers to determine quadrat positions. We
determined the position of each quadrat by taking two
random numbers (x and y coordinates), one between 0.0
and 5.9 (0.1 increments), which set the position of the
quadrat along the measuring tape axis, and a second
between 0 and 4, which set quadrat position perpendicular to the measuring tape (0 /two squares below the
measuring tape; 1 /immediately below; 2 /immediately
above; 3 /two squares above). There were thus a total of
240 possible locations where quadrats could be placed at
each intertidal level. We thoroughly searched each
quadrat, picking up and inspecting every cobble and
oyster shell at the lower intertidal levels and cutting off
Fucus parts at higher levels, and collected all snails
greater than 2 mm in shell length. We sampled until we
had a minimum of 50 snails at each intertidal level, and a
minimum of 15 different quadrats had been taken. The
maximum number of quadrats taken at any intertidal
level was 40. All snails were brought back to the
laboratory, identified under a dissecting microscope,
and measured (maximum shell length) to the nearest
0.01 mm using digital calipers.
We used chi-square analyses of contingency tables to
test two hypotheses: (1) L. subrotundata and L. sitkana
are randomly, or uniformly, distributed among tidal
levels and (2) L. subrotundata and L. sitkana have
similar vertical distributions. We did separate analyses
for the two species (hypothesis 1) and sites (hypotheses 1
and 2). We then determined whether snail size varied
among intertidal heights. We first tested for homogeneity
of variances among groups using Bartlett’s test (we did
not use Cochran’s test because sample sizes were
unequal), analyzing different species and sites separately.
Data for L. sitkana at both sites, and for L. subrotundata
at site B, passed the homoscedasticity test (P /0.05),
and were analyzed with ANOVA’s followed by Tukeytype multiple comparisons. Data for L . subrotundata at
site A did not pass the test (P B/0.01), and were analyzed
with a Mann-Whitney U-test (L. subrotundata was only
found at 2.4 and 2.8 m) using the normal approximation
for statistical testing.
Experiment 1: quantification of predation risk
We conducted tethering experiments at our two study
sites, between 9 and 18 August 2000, to compare
predation risk faced by L. sitkana and L. subrotundata
at low (i.e. 1.2 m) and high (i.e. 2.5 m) intertidal levels.
Based on results obtained recently by Rochette and Dill
(2000), we selected these two intertidal heights for the
515
experiment because (1) they span nearly the entire
vertical range of littorinids ( :/1 /3 m), (2) littorinid
density typically peaks around these heights, and (3)
predation risk faced by littorinids varies significantly
between these levels.
For this experiment, snail sizes were chosen to reflect
naturally occurring differences in adult body size between species and intertidal levels. Although larger snails
experience a higher risk of predation than small snails
(McCormack 1982, Rochette and Dill 2000) and L.
sitkana typically reaches larger body sizes than L.
subrotundata (R. Rochette, pers. obs.), we did not use
similar-size L. sitkana and L. subrotundata because we
wanted to mimic the actual mortality rates experienced
by snails in nature. We therefore used random quadrat
sampling (as described for the vertical distribution
surveys) to collect snails appearing to be larger than 4
mm in shell length. Because L. sitkana showed a wide
vertical distribution at both study sites (Results) and
adults were larger at the 2.8- than at the 1.2-m level
(Results), half of the L. sitkana snails used in the
tethering experiments were collected from the 1.2-m
level and the other half from the 2.8-m level. In contrast,
we randomly collected all L. subrotundata snails from
the 2.8-m level, because this species had a much
narrower vertical distribution and did not show consistent size variation between intertidal levels (Results).
In the laboratory, we measured all snails collected to
the nearest 0.01 mm with digital calipers, lined up
individuals of a given species and origin (site and
intertidal height), and used a random-number generator
to determine which snails would be used in the tethering
experiment and what position they would occupy on the
transect (below). We then thoroughly air-dried each
snail’s shell, attached a 10 cm piece of 2.25 kg test
monofilament (diameter /130 mm) to the apex of each
shell with epoxy (we made a small knot in the line to
increase adhesion) and let the glue dry for approximately
30 min. Each snail was then placed in a separate, labeled
eppendorf tube. We then half-filled each tube with fresh
seawater and placed them in running seawater overnight.
The following morning, we attached 10 snails of each
species (5 L. sitkana ’s from the 1.2-m level and 5 from
the 2.8-m level; 10 L . subrotundata ’s from the 2.8-m
level) to each of four 6-m long transects (2 sites /2
intertidal heights) made of 25 kg test monofilament
(diameter /730 mm); we randomly determined the order
of each snail on a given transect, and left 30 cm between
snails to minimize interactions between individuals.
We recorded the fate of individual snails at 3-d
intervals over a 9-d period (i.e. 3 temporal replicates),
replacing dead or missing snails on each occasion with
individuals of the same species (and origin, in the case of
L. sitkana ). For each position on the transect we
recorded whether the snail was (1) alive (snail visible
and responded to touch), (2) crushed (when only a
516
broken shell or shell fragments were found imbedded in
the epoxy), (3) empty (when the shell was intact but
without a snail), and (4) missing (when only a monofilament knot was recovered, or a piece of epoxy with no
visible shell fragments). We estimated predation rates by
calculating, for each temporal replicate and combination
of factor levels (i.e. species, site and intertidal height), the
number of snails that were recovered as crushed, missing
or empty, and then dividing that quantity by the number
of snails that had been released in that particular
treatment (usually 10 snails, but occasionally 9 or 8
due to losses). In other words, we assumed that all snails
not recovered alive had been killed by predators. We
made this assumption because the recent study by
Rochette and Dill (2000) indicated that tethered snails
that are deployed in Bamfield inlet under predator-proof
cages are virtually never recovered as ‘‘missing’’ or
‘‘empty’’.
We analyzed mortality rates with a 3-way factorial
ANOVA, using site (A and B), intertidal height (1.2 and
2.5 m) and species (L. sitkana and L. subrotundata ) as
fixed-effect factors. We considered site to be a fixedeffect factor, as opposed to a random-effect factor,
because we wanted to assess the consistency of effects
across our two sites; we were not simply trying to control
for site-dependent variability, but rather wanted to test
statistically whether effects were the same at the two
sites. The raw data was heteroscedastic (Cochrans’
C8,2 /0.684; PB/0.01), because variances tended to
increase with mean mortality rates. However, a square
root transformation (i.e. sqrt(x)/sqrt(x/1)) satisfyingly
stabilized the variances (Cochrans’ C8,2 /0.445; P/
0.05). We used Tukey-type multiple comparisons
(Zar 1984) and a ‘‘family-wise’’ error rate of 0.05 to
interpret significant interaction terms. The critical value
for all comparisons was q0.05,16,2 / 2.998 (16 corresponds to the number of degrees of freedom associated
with the model error term, and 2 corresponds to the
number of means involved in each family of comparisons).
We also performed a similar factorial ANOVA to
determine if there were differences in mortality rates
between high-origin and low-origin L. sitkana snails.
Experiment 2: quantification of growth rates
We quantified growth rates at our two study sites
between February 22 and March 15, 2001. Using
random quadrat sampling (as described for the vertical
distribution surveys), we collected approximately 200
small (less than 4 mm shell length) L. sitkana from each
intertidal height at each site. In contrast (due to their
narrow vertical distribution), we randomly collected
similar numbers of small L. subrotundata at each site,
but only from the high-intertidal level. We also collected
OIKOS 105:3 (2004)
substrate (small rocks and pieces of [chiseled] bedrock
with and without Fucus attached) from each site/height
combination.
Upon returning to the lab, we measured all snails to
the nearest 0.01 mm, entered the data on a computer,
and used a random-number generator (snails were
aligned on paper towels) to create 16 groups of snails
per site (i.e. eight L. sitkana groups and eight L.
subrotundata groups per site). Each group consisted of
ten individuals of the same species; in the case of L.
sitkana , groups were composed of five high-origin and
five low-origin individuals. Snails within each group were
then individually color-marked. We used five different
colors of enamel-based paint (white, yellow, red, blue
and green) and two mark locations (body whorl only and
body whorl/apex) to created 10 distinct marks. Snails
were marked under the dissecting microscope using a
single paintbrush hair, then left to dry for 30 min.
Each group of snails was placed in a small, plasticframed cage (13 /13/5 cm), with walls of 500 micron
Nytex. Substrate from the appropriate site and intertidal
height was added to each cage. Lids were hot-glued to
each cage to prevent snails from escaping during the
subsequent growth period. Cages were then tethered to
bricks and placed in the field at the appropriate site and
intertidal height. Thus, at each site, there were four
replicate cages of snails of each species at each intertidal
height. Cages were recovered on March 15, 2001, at
which time we counted the number of snails still alive,
and measured them to the nearest 0.01 mm.
We analyzed survivorship of snails during the growth
experiment with a 3-way factorial ANOVA, using site (A
and B), intertidal height (1.2 and 2.5 m) and species (L.
subrotundata and L. sitkana ) as fixed-effect factors. We
then used hierarchical (i.e. nested) factorial ANOVAs to
analyze the growth of snails remaining alive; sites,
heights and species were again used as fixed effect
factors, and cages were nested within these three factors.
Significance testing for the main effects (and their
interactions) was done using the nested variable (i.e.
cage) as the error term (i.e. the MS for the nested factor
was used as denominator of the F-ratios). We conducted
three separate hierarchical ANOVAs, because highorigin and low-origin L. sitkana snails displayed different growth rates (F1,12 /7.99, PB/0.05). In the first
analysis, we ‘‘pooled’’ L. sitkana snails originating from
high- or low-intertidal areas, and tested for overall
growth differences between species. In the second set of
analyses, we compared growth rates of L. subrotundata
to high-origin or low-origin L. sitkana snails, separately.
In the case of the comparison between L. subrotundata
and low-origin L. sitkana , the raw data were heteroscedastic (Cochrans’ C8,3 /0.450; PB/0.05), but a
square root transformation (i.e. sqrt(x)/sqrt(x/1))
satisfyingly stabilized the variances (Cochrans’ C8,3 /
0.346; P/0.05). In all of the above analyses, we used
OIKOS 105:3 (2004)
Tukey-type multiple comparisons and a ‘‘family-wise’’
error rate of 0.05 to interpret significant interaction
terms (Zar 1984). The critical value for these multiple
comparisons was q0.05,24,2 / 2.919 (24 corresponds to
the number of degrees of freedom associated with the
nested [cage] factor, and 2 corresponds to the number of
means involved in each family of comparisons).
We did not use initial length as a covariate in these
analyses, because it was weakly and inconsistently
related to snail growth; Pearson product-moment correlations indicated a significant (P B/0.05) positive relation
between initial length and growth in only two of the 12
different treatment combinations. The absence of a
consistent relationship between initial snail size and
growth is not surprising considering the narrow size
range of snails we used (i.e. 100% and 80% of individuals
within a range of 1.0 mm and 0.6 mm, respectively).
Furthermore, a hierarchical factorial ANOVA (above)
indicated that snail shell length was uniform across
treatments at the beginning of the experiment (all main
effects P/ 0.6).
Results
Snail spatial distribution
As expected, L. sitkana and L. subrotundata distributions were not random with respect to intertidal level,
and each species predominated in a different part of the
intertidal zone (Fig. 1). At both sites, the number of L.
sitkana (site A: x2 /78.8, DF/4, P B/0.0001; site B:
x2 /40.1, DF /4; P B/0.0001) and L. subrotundata (site
A: x2 /184, DF /4, P/0.0001; site B: x2 /79, DF/4,
PB/0.0001) snails at different intertidal heights differed
markedly from numbers expected based on the number
of quadrats taken and assuming a random, or uniform,
vertical distribution. But more relevant to our current
Fig. 1. Observed densities (snails per m2; x̄/1 SE) of L. sitkana
and L. subrotundata at various intertidal heights at each of our
two study sites.
517
hypothesis, L. sitkana and L. subrotundata showed
nearly opposite distribution patterns at both of our
study sites (site A: x2 /216.1, DF/4, P/0.0001; site
B: x2 /198.2, DF/4, P /0.0001), with L. sitkana
predominating in lower portions of the intertidal zone
and L. subrotundata occurring only at the two highest
levels sampled (Fig. 1).
Although not shown here, we also searched for snails
below (i.e. 0.8 m) and above (i.e. 3.2 m) our study sites.
However, very few were found and the identity of those
that we did find was consistent with the distribution
patterns presented above. Thus, at both sites small
numbers of L. sitkana were found at the 0.8-m level,
but no L. subrotundata . At site A, many snails were
found underneath three rocks (each about 20 cm in
diameter) sitting on bare bedrock at the 3.2-m level; all
were L. subrotundata . No snails of either species were
found at the 3.2-m level of site B, which was shaded and
overgrown by the green algae Enteromorpha intestinalis.
In L. sitkana , snail size varied among intertidal
heights at both study sites (Fig. 2; site A: F4,194 /
26.34, P B/0.0001, site B: F4,197 /62.59; PB/0.0001).
Multiple comparisons indicated that, at site A, L.
sitkana were significantly larger at the highest intertidal
level (i.e. 2.8 m) than at all other levels (P B/0.05), and
that snail size was similar from 1.2 to 2.4 m (P /0.05).
At site B, snails at the three middle levels (1.4, 1.8 and
2.2 m) were of similar size (P /0.05), but significantly
larger than snails at the lowest level (P B/0.05), and
smaller than snails at the highest level (P B/0.05). This
vertical size gradient of L. sitkana snails (McCormack
1982) is likely due to larger individuals experiencing high
mortality rates in low-intertidal areas (McCormack
1982, Behrens Yamada and Boulding 1996, Rochette
and Dill 2000), and to differences in size at sexual
maturity between high- and low-shore snails (Rochette et
al. 2003). In L. subrotundata , snails collected at the 2.8 m
level of site A were significantly larger than those from
the 2.4 m level (Z / /3.231, P/0.001). This difference
is of dubious ecological significance, however, because
only 6 snails were found at the 2.4-m level; the density
estimates indicate that less than 5% of the population
occurred at that level. At site B, the size of L.
subrotundata did not differ between intertidal levels
(F1,60 /2.47; P/0.121).
Thus, in general, snail size tended to increase with
increasing intertidal height. Furthermore, as expected
from anecdotal observations (Table 1 in Behrens Yamada 1992), L. sitkana and L. subrotundata tend to be
spatially segregated at our study sites, with L. sitkana
occurring primarily at low intertidal heights and L.
subrotundata exclusively at higher intertidal heights.
Experiment 1: quantification of predation risk
We made a total of 232 observations on tethered snails,
51 of which (22%) were classified as predation events;
78% of the snails assumed to have been killed by
predators were recovered as ‘‘crushed’’, 16% as ‘‘missing’’ and 6% as ‘‘empty’’ (as described in Methods). The
factorial ANOVA revealed two significant interactions
between main effects (Table 2; Fig. 3). First, and most
importantly, there was a significant interaction between
snail species and intertidal height (P /0.001), with
variation in mortality rates between low- and highintertidal areas seemingly greater for L. subrotundata
than for L. sitkana (Fig. 3). The multiple comparisons
indicated that at high-intertidal levels, both species had
similar mortality rates (q16,2 /2.411; P/0.05), but at
low-intertidal levels L. subrotundata was killed more
frequently than L. sitkana (q16,2 /5.451; PB/0.05). For
both species, however, mortality rates were much greater
in low- than in high-intertidal areas (L. sitkana : q16,2 /
11.248, P/0.05; L. subrotundata : q16,2 /19.110, P/
0.05). The second significant interaction was between
site and intertidal height (P/0.023). This weak interaction appeared due to mortality rates being slightly
greater at site A (versus B) in the high-intertidal, and at
site B (versus A) in the low intertidal, although the
difference in mortality between sites was not significant
Table 2. Tethering experiment. Factorial ANOVA comparing
mortality rates among snail species, intertidal heights and study
sites.
Fig. 2. Shell lengths (mm; x̄9/1 SE) of L. sitkana and L.
subrotundata at various intertidal heights at each of our two
study sites.
518
Source of variation
DF SS ( /10 1)
Species (Sp)
Intertidal height (H)
Site (S)
Sp/H
Sp/S
H/S
Sp/H/S
Error
1
1
1
1
1
1
1
16
0.353
35.212
0.002
2.362
0.360
0.973
0.127
2.445
F
P
2.311
230.396
0.013
15.453
2.356
6.365
0.768
0.148
/0.001
0.912
0.001
0.144
0.023
0.394
OIKOS 105:3 (2004)
species at the low- (P B/0.05) but not at the high- (P /
0.05) intertidal level.
Thus, regardless of which analysis is used, both species
agree on the relative ranking of habitats with respect to
mortality risk. However, L. subrotundata appeared to
experience a higher ratio of mortality risk across habitats
than L. sitkana .
Experiment 2: quantification of growth rates
Fig. 3. Tethering experiment. Overall mortality rates (%; x̄/1
SE) of L. sitkana and L. subrotundata in the high- and lowintertidal zone at each of our two study sites.
at either height (high-intertidal: q16,2 /2.411, P /0.05;
low intertidal: q16,2 /2.635, P/0.05).
Earlier studies have found that larger snails experience
a higher risk of predation than small snails in Bamfield
inlet (McCormack 1982, Rochette and Dill 2000).
However, size did not appear to be a significant factor
during this experiment, because we found no evidence of
differences in mortality rate between L. sitkana snails
originating from high-intertidal and low-intertidal areas
(P/0.51), despite the fact that the former were significantly larger than the latter (Fig. 2). L. subrotundata
snails were larger than low-origin L. sitkana snails, but
smaller than high-origin ones (Fig. 2).
Due to time constraints, snails that were recovered as
‘‘alive’’ were not replaced between temporal replicates; it
would have been difficult to replace all snails during low
tide at the end of a tidal sequence, in particular, when
low-intertidal transects were only exposed for 1 /2 h.
Thus, the above analysis can be criticized on the grounds
that data points are not truly independent. In order to
address this criticism, we pooled the three temporal
replicates and re-analyzed the data. In this case, percent
mortality was analyzed by 2-way ANOVA, using snail
species and intertidal height as factors and the two study
sites as independent replicates. This analysis revealed a
marginally significant interaction between snail species
and intertidal height (P /0.09), and multiple comparisons indicated that mortality rates differed between
OIKOS 105:3 (2004)
Seventy three percent of the snails survived the field
growth experiment, but patterns of survivorship differed
between species (Table 3, Fig. 4). First, and most
importantly, the factorial ANOVA revealed a significant
interaction between snail species and intertidal height
(P/0.002). The multiple comparisons indicated that at
high-intertidal levels, both species survived similarly well
(q24,2 /0.289; P /0.05), but at low-intertidal levels
survivorship was much greater for L. sitkana than for
L. subrotundata (q24,2 / 7.217; P /0.05). Further, and
perhaps more importantly, survivorship of L. sitkana
was independent of intertidal height (q24,2 /1.732; P/
0.05), but that of L. subrotundata was significantly
greater in high- than in low-intertidal areas (q24,2 /
5.200; P/0.05). The second significant interaction was
between site and species (P /0.008). Survivorship of L.
sitkana was greater at site B than A (q24,2 /4.619; PB/
0.05), whereas that of L. subrotundata was similar at
both sites (q24,2 /1.155; P/0.05). Further, survivorship
was similar for both species at site A (q24,2 /0.866; P/
0.05), but significantly less for L. subrotundata than for
L. sitkana at site B (q24,2 /6.640; P/0.05).
The hierarchical factorial ANOVAs indicated similarities, as well as differences, between growth patterns of
L. subrotundata and L. sitkana snails (Table 4 /6, Fig.
5). The first analysis revealed a significant interaction
(P/0.002) between snail species and intertidal height
(Table 4, Fig. 5A), which indicated that the effect of
intertidal height on growth was not the same for the two
species. Multiple comparison tests indicated that both
species grew more in the low intertidal than in the highintertidal (L. sitkana : q24,2 /11.860, P/0.05; L. subrotundata : q24,2 / 3.843, P B/0.05), but the magnitude of
Table 3. Growth experiment. Factorial ANOVA comparing
snail survivorship among species, intertidal heights and study
sites.
Source of variation
DF
SS
F
P
Species (Sp)
Intertidal height (H)
Site (S)
Sp/H
Sp/S
H/S
Sp/H/S
Error
1
1
1
1
1
1
1
24
0.211
0.045
0.045
0.180
0.125
0.001
0.001
0.360
14.083
3.000
3.000
12.000
8.333
0.083
0.083
0.001
0.096
0.096
0.002
0.008
0.775
0.775
519
Table 6. Growth experiment. Hierarchical factorial ANOVA
comparing growth of L . subrotundata and low-origin L .
sitkana in the high and low-intertidal zone. To satisfy the
homoscedasticity assumption, this analysis was performed on
square-root transformed data. Significance testing for the main
effects (and their interactions) was done using the cage factor as
the error term (i.e. the MS for the nested factor was used as
denominator of the F ratios).
Source
DF
SS
F
Intertidal height (IH)
Species (Sp)
Site (Si)
IH/Sp
IH/Si
Sp/Si
IH/Sp/Si
Cage [Sp, Si, IH]
Error
1
1
1
1
1
1
1
24
139
1.302
0.114
0.040
0.184
0.009
0.105
0.022
1.174
6.056
26.622
2.337
0.812
3.771
0.176
2.136
0.388
1.122
P
/0.001
0.139
0.377
0.064
0.679
0.157
0.539
0.328
Fig. 4. Growth experiment. Survivorship (%; x̄/1 SE) of L.
sitkana and L. subrotundata snails caged in the high- and lowintertidal zone at each of our two study sites.
Table 4. Growth experiment. Hierarchical factorial ANOVA
comparing growth of L . subrotundata and L . sitkana in the
high and low-intertidal zone. Significance testing for the main
effects (and their interactions) was done using the cage factor as
the error term (i.e. the MS for the nested factor was used as
denominator of the F ratios).
Source
DF
SS
F
P
Intertidal height (IH)
Species (Sp)
Site (Si)
IH /Sp
IH /Si
Sp/Si
IH /Sp/Si
Cage [Sp, Si, IH]
Error
1
1
1
1
1
1
1
24
203
2.229
0.028
0.001
0.495
0.019
0.042
0.021
0.941
6.837
56.858
0.701
0.031
12.635
0.484
1.071
0.531
1.164
/0.001
0.411
0.862
0.002
0.493
0.311
0.473
0.279
Table 5. Growth experiment. Hierarchical factorial ANOVA
comparing growth of L . subrotundata and high-origin L .
sitkana in the high and low-intertidal zone. Significance testing
for the main effects (and their interactions) was done using the
cage factor as the error term (i.e. the MS for the nested factor
was used as denominator of the F ratios).
Source
DF
SS
F
Intertidal height (IH)
Species (Sp)
Site (Si)
IH /Sp
IH /Si
Sp/Si
IH /Sp/Si
Cage [Sp, Si, IH]
Error
1
1
1
1
1
1
1
24
136
1.859
0.196
0.008
0.514
0.031
0.003
0.003
0.517
4.273
86.223
9.082
0.367
23.830
1.418
0.118
0.147
0.686
520
P
/0.001
0.006
0.550
/0.001
0.245
0.734
0.705
0.858
Fig. 5. Growth experiment. Growth (mm; x̄/1 SE) of L.
subrotundata and L. sitkana in the high- and low-intertidal
zone at each of our two study sites; (A) high- and low-origin L.
sitkana pooled, (B) high- and low-origin L. sitkana plotted
separately. Note that non-transformed values are shown
throughout, even though the comparison between L. subrotundata and low-origin L. sitkana was done on square root
transformed data.
this growth difference between intertidal heights appeared greater for L. sitkana than for L. subrotundata
(Fig. 5A). This latter interpretation is supported by the
fact that L. sitkana grew significantly more than L.
subrotundata in the low intertidal (q24,2 /4.328, PB/
0.05), but not in the high-intertidal (q24,2 /2.901, P/
0.05). In fact, the exact opposite pattern appeared true in
the high-intertidal (L. subrotundata growing more
than L. sitkana ), but this species effect was not
statistically significant (recall that q0.05,24,2 / 2.919).
There was no significant difference in growth rates
OIKOS 105:3 (2004)
among cages belonging to the same treatment (P /
0.279).
The analysis involving L. subrotundata and highorigin L. sitkana snails also revealed a highly significant
(P B/ 0.001) interaction between species and intertidal
height (Table 5, Fig. 5B). Multiple comparison tests
indicated that both species grew more in the low
intertidal than in the high-intertidal (L. sitkana :
q24,2 /13.492, P/0.05; L. subrotundata : q24,2 /5.289,
P B/0.05), but the magnitude of this growth difference
between intertidal heights appeared greater for L.
sitkana than for L. subrotundata (Fig. 5B). This latter
interpretation is supported by the fact that L. sitkana
grew significantly more than L. subrotundata in the lowintertidal (q24,2 /8.293, P /0.05), where growth rates
were greater, but not in the high-intertidal (q24,2 /1.981,
P /0.05), were growth rates were lower. There was no
significant difference in growth rates among cages
belonging to the same treatment (P /0.858).
The analysis involving L. subrotundata and low-origin
L. sitkana snails also revealed a highly significant
intertidal-height effect (P /0.001); snails grew more in
low- than high-intertidal areas (Table 6, Fig. 5B). Again,
this effect seemed more pronounced for L. sitkana than
for L. subrotundata (Fig. 5B), although the interaction
between species and intertidal height only approached
significance (Table 6, P/0.064). There was no significant difference in growth rates among cages belonging to
the same treatment (P /0.328).
In summary, both species grew more rapidly in the
lower than the upper parts of the intertidal zone, but the
growth differential between high- and low-intertidal
habitats is greater for L. sitkana than for L. subrotundata . These conclusions seem to hold whether we
consider L. sitkana snails originating from the low- or
the high-intertidal zone, although the latter grew faster
than the former in all habitats (F1,12 /7.99, PB/0.05).
Discussion
As suggested by earlier, anecdotal observations, L.
sitkana and L. subrotundata tend to be segregated across
intertidal heights at our study sites, with L. sitkana
occurring predominantly in mid- to high-intertidal
( /1 /2.8 m) habitats and L. subrotundata exclusively
in high- to extra-high ( /2.4 /3.2 m) intertidal habitats
(Behrens Yamada 1992). More extensive surveys have
revealed that this pattern is very consistent within
Bamfield inlet (R. Rochette, unpubl.). According to the
results of this study, both species experience higher
predation and higher growth in low- than in highintertidal habitats, and hence, agree on the rankings of
habitat riskiness and productivity. Thus, both Grand and
Dill’s (1999) ‘‘differential risk ratios’’ hypothesis and
Brown’s (1998) ‘‘interspecific trade-offs’’ hypothesis
OIKOS 105:3 (2004)
(Table 1) provide potential explanations for the observed
spatial distribution of this pair of coexisting species.
Although both species agreed on ranking of habitats
in terms of growth potential and mortality risk, L.
sitkana and L. subrotundata experienced differences in
their habitat-specific growth rates and mortality risks.
Despite being similarly at risk of predation in highintertidal habitats, they did not experience equal mortality in low-intertidal habitats. L. subrotundata was
subject to significantly higher mortality than L. sitkana
at the low-intertidal height. It is worth noting that this
pattern was not only found with the tethering experiment, but also during the growth experiment, in which
snails were protected from predators inside cages. This
finding indicates that predators and some other factor(s)
(e.g. grazing efficiency) more negatively impact the
survivorship of L. subrotundata than L. sitkana in lowintertidal areas. Thus, L. subrotundata perceived a higher
ratio of mortality (i.e. ‘‘risk ratio’’; Grand and Dill 1999)
across the two habitats than L. sitkana (regardless of the
intertidal height of origin of the latter). In contrast,
growth rate differences between habitats were greater for
L. sitkana than for L. subrotundata . Whereas both
species grew at the same rate at the high-intertidal level,
L. sitkana individuals grew more rapidly than L.
subrotundata snails at the low-intertidal level. This
interspecific difference in habitat-dependent growth
differential seemed to hold whether we considered
high-origin or low-origin L . sitkana snails (Fig. 5),
although the species by height interaction term was only
marginally significant (P /0.06) in the latter case.
According to Brown’s ‘interspecific trade-off’ hypothesis, the observed pattern of habitat selection (and
consequently, coexistence of the two species) is predicted
to occur when L. subrotundata is more vulnerable to
predation in both low- and high-intertidal habitats and
L. sitkana better at acquiring resources in both habitats.
Although L. subrotundata was more vulnerable to
predators than L. sitkana in the higher risk, lowintertidal habitat, mortality rates did not differ between
species in the high-intertidal habitat. Similarly, depending on the intertidal height of origin of L. sitkana ,
species growth rates appeared to differ at either high- (L.
subrotundata /low-origin L. sitkana ; Fig. 5B) or lowintertidal levels (high-origin L. sitkana /L. subrotundata ), but not at both simultaneously. Thus, assuming
that our experiment had sufficient power to detect such
differences if present, the observed pattern of habitat
selection cannot be solely attributed to interspecific
trade-offs in competitive ability and vulnerability to
predation. It should also be noted that this hypothesis
(like Brown’s 1998 two alternative hypotheses) assumes
equal resource availability in the two habitats. It is
unclear how the model’s predictions might change if
habitats differ in the quantity of resources they provide.
521
In contrast, Grand and Dill’s (1999) ‘‘risk ratios’’
hypothesis does not require that each species be superior
at a single task (i.e. competing for resources or avoiding
predators) in both habitats for segregation across
habitats, and consequently, coexistence (Grand 2002)
to occur. Instead, the observed pattern of habitat
segregation is predicted to occur when L. subrotundata
experiences a higher ratio of mortality risk across the
habitats than L. sitkana , regardless of which species is
absolutely at greater risk of predation in the riskier, lowintertidal habitat. However, central to Grand and Dill’s
(1999) hypothesis is the assumption that the competitive
abilities (i.e. the ability to capture and consume resources) of each species remain constant across habitats.
Assuming that growth rate accurately reflects competitive ability, our growth experiment suggests that relative
competitive abilities of the two species are not independent of habitat or for L. sitkana , habitat of origin.
However, it is unclear whether violating this underlying
assumption of Grand and Dill’s (1999) model should
lead to a pattern of habitat selection other than
segregation of competing species across habitats. Indeed,
one might expect segregation to be even more complete
when each species is the better competitor in the habitat
in which it predominates (MacArthur and Levins 1967,
Lawlor and Maynard Smith 1976, Vincent et al. 1996),
despite the additional complication of habitat differences
in mortality risk. We are currently working on a model
that addresses situations where interspecific differences
in competitive abilities and vulnerability to predation
both change across habitats.
Taken together, our results are most consistent with
the mechanism of spatial segregation proposed by
Grand and Dill (1999). L. sitkana and L. subrotundata
tend to be segregated intertidally at our study sites not
because each is superior at competing for resources or
avoiding predators in different habitats, but because
differences between species in habitat-specific predation
risks result in L. subrotundata experiencing a higher
ratio of mortality risk across habitats. The actual
mechanism leading to this segregation pattern could be
based on competitive exclusion or differences in behavioral preferences, or a combination of both processes.
One important assumption of our study is that the
tethering experiment adequately reflects differences in
relative mortality risk between species and intertidal
heights. In other words, if snail mortality rates are
affected by the tethering procedure, the magnitude of
this bias (Zimmer-Faust et al. 1994) must be constant, or
additive, among groups to be compared (Peterson and
Black 1994). Rochette and Dill (2000) recently tested this
assumption for L. sitkana and L. scutulata snails near
our study sites using various experimental procedures
(e.g. mark-release-recapture experiments, varying lengths
of tethers, predator-proof cages) and they found no
522
evidence that tether biases were not constant across snail
species, sizes or intertidal heights. We are therefore
confident that our mortality estimates can be used to
compare mortality risk of our different snail groups. In
fact, not only are our tethering results reliable in relative
terms, they probably even reflect absolute rates of snail
mortality. Indeed, in their study Rochette and Dill (2000)
reported similar rates of mortality for tethered and nontethered snails. This lack of a tether bias should perhaps
not come as a surprise, because snails can not outrun or
outmaneuver their main predators (i.e. crabs and fishes)
in these habitats (Barbeau and Scheibling 1994).
One may wonder how snail populations could be
maintained under such high mortality rates (i.e. between
:/30 /60% of snails killed in 3 d in the low intertidal),
but it must be noted that predation rates are not always
this high; they drop to :/5 /10% snails killed in 3 d
during fall and winter months (Rochette and Dill,
unpubl.). Furthermore, females are highly fecund, producing :/25 /175 eggs (depending on their size) per
single spawning event, and laying several egg masses per
year. Thus, although we do not possess the demographic
data (e.g. annual fecundity is not known) to confirm that
snail populations can sustain themselves under the
mortality rates estimated using tethering, this appears
quite probable.
Clearly, the high- and low-intertidal habitats described
in this study pose different ecological challenges to the
littorinid snails that inhabit them. Although littorinids
are highly mobile animals, displaying oriented movement patterns in response to unfavorable conditions
(Rochette and Dill 2000), the distribution patterns
documented in this study are unlikely to result from
immigration of snails from outside the study area, as
they are not found above or below the intertidal range
covered in our survey, nor are they likely to migrate from
wave-exposed headlands many kilometers away. It also
seems unlikely that snails would routinely move between
upper and lower-intertidal areas to sample changes in
environment conditions. In our study, individuals were
physically prevented from moving between habitats in
response to experimental manipulations, suggesting that
the observed habitat-specific differences in competitive
ability and vulnerability to predation reflect local
adaptation over evolutionary time. Given the ongoing
pressures faced by these two species, continued spatial
segregation is likely to lead to further interspecific
differences in traits related to competitive ability and
vulnerability to predation. Indeed, an extension of
Grand’s (2002) model over evolutionary time scales
suggests that coexisting pairs of species, such as L.
sitkana and L. subrotundata , must continue to diverge in
their abilities to compete for resources and/or escape
predation for continued coexistence to occur (T. Grand,
unpubl.). Interestingly, electrophoretic studies of enzyme
OIKOS 105:3 (2004)
variations at the leucine-aminopeptidase locus suggest
that natural selection may contribute to furthering niche
segregation among sympatric and ecologically similar
species of limpets (Murphy 1976).
The zonation pattern of intertidal organisms is
undoubtedly affected by many biotic and abiotic factors
operating in concert over both ecological and evolutionary time scales. Different studies have examined
zonation of littorinid snails in light of interspecific
differences in metabolic performance and physiological
tolerance (Sacchi 1969, Sokolova et al. 2000, Sokolova
and Pörtner 2001), feeding preferences and specialization (Sacchi 1969, Sacchi and Voltolina 1987), and
susceptibility to predation (Elner and Raffaelli 1980).
Over ecological time scales, these factors directly affect
critical biological processes such as survival, growth,
reproduction and (in the case of motile organisms)
movement. Over evolutionary time scales, they select
phenotypes that are better adapted to conditions prevailing in different parts of the intertidal zone. To the
best of our knowledge, our study is the first to analyze
niche partitioning among intertidal snails in relation to
interspecific differences in trade-offs involving feeding
potential and predation risk. We believe that the greater
vulnerability of L. subrotundata in high-risk areas
compared to L. sitkana is due to its shell being less
resistant to crab predation. The mineralogy of both
species’ shell is similar (Reid 1996), but the shell of L.
subrotundata is lighter (i.e. contains less calcified material) than that of a similar-sized L. sitkana (Boulding et
al. 1999).
We hypothesize that vertical segregation between L.
subrotundata and L. sitkana is further facilitated by their
mode of development. In comparison to pelagic development, benthic larval development enhances the predictability of recruitment patterns (Underwood 1979).
Furthermore, benthic development may reduce gene
flow among individuals inhabiting different habitats,
which should favor local adaptation and could therefore
reduce niche overlap among species (above). This
potential for localized adaptation is supported by the
differences in growth rates we observed between highorigin and low-origin L. sitkana . These phenotypic
differences appear to reflect life-history adaptation to
spatial and size-dependent predation patterns in the
intertidal zone (Rochette et al. 2003). More specifically,
because mortality risk is greater in the lower intertidal
and on larger snails, low-intertidal individuals grow
slower but become sexually mature at a smaller size,
whereas high-intertidal ones grow more quickly, but
mature at a larger body size. Recent work suggests that
these phenotypic differences have at least a partial
genetic basis (Rochette and Dill, unpubl.). In Europe,
another littorinid snail that lacks a long range dispersing
larvae, L. saxatilis (Olivi), displays extensive phenotypic
OIKOS 105:3 (2004)
variation between high- and low-intertidal areas, and
much of this variation is under genetic control (Johannesson et al. 1993, 1995, 1997, Rolán-Alvarez et al.
1996).
Although much of the published competitioncoexistence literature emphasizes the importance and
prevalence of interspecific trade-offs in competitive
ability and vulnerability to predation, such trade-offs
may not be ubiquitous, in particular, when competitor
species are morphologically similar, and habitats have
similar types of resources and house the same species of
predators. Under such circumstances, individual behavioural models are likely to provide explanations for
coexistence that traditional, population models cannot.
Furthermore, if behavioural diversification precedes
morphological diversification (Wcislo 1989, McLaughlin
et al. 1999), specialization of species on different
resources or habitats may be an evolutionary consequence of differential habitat selection rather than a
cause. Given the current interest in linking adaptive
individual behavior to ecological patterns (Sutherland
1996, Brown 1998), illuminating the conditions under
which morphologically similar species are likely to
engage in different behaviors should be of general
importance to both behavioural and evolutionary ecologists.
Acknowledgements / The research was funded by NSERC
operating grant A6869 to L. M. Dill (Simon Fraser University)
and a WCUMBS Research Associate Award (Bamfield Marine
Sciences Centre, British Columbia) to RR. TCG thanks B.
Crespi for toddler-minding during the collection of mortality
risk data and an NSERC postdoctoral fellowship for financial
support. RR gratefully acknowledges support from an NSERC
postdoctoral fellowship and a WCUMBS Research Associate
Award.
References
Barbeau, M. A. and Scheibling, R. E. 1994. Procedural effects
of prey tethering experiments: predation of juvenile scallops
by crabs and seastars. / Mar. Ecol. Prog. Ser. 111: 305 /310.
Behrens Yamada, S. 1992. Niche relationships in northeastern
Pacific littorines. / In: Grahame J., Mill P. J. and Reid D. J.
(eds), Proc. Third Int. Symp. on Littorinid Biology. 281 /
291.
Behrens Yamada, S. and Boulding, E. G. 1996. The role of
highly mobile crab predators in the intertidal zonation
of their gastropod prey. / J. Exp. Mar. Biol. Ecol. 204:
59 /83.
Boulding, E. G., Holst, M. and Pilon, V. 1999. Changes in
selection on gastropod shell size and thickness with waveexposure on northeastern Pacific shores. / J. Exp. Mar. Biol.
Ecol 232: 217 /239.
Brown, J. S. 1998. Game theory and habitat selection. /In:
Dugatkin, L. A. and Reeve, H. K. (eds), Game theory and
animal behavior. Oxford Univ. Press, pp. 188 /220.
Elner, R. W. and Raffaelli, D. G. 1980. Interactions between two
marine snails, Littorina rudis Maton and Littorina nigrolineata Gray, a predator, Carcinus maenas (L.) and a parasite,
Microphallus similes Jägerskiold. / J. Exp. Mar. Biol. Ecol
43: 151 /160.
523
Grand, T. C. 2002. Foraging-predation risk trade-offs, habitat
selection, and the coexistence of competitors. / Am. Nat.
159: 106 /112.
Grand, T. C. and Dill, L. M. 1999. Predation risk, unequal
competitors, and the ideal free distribution. / Evol. Ecol. 1:
331 /347.
Johannesson, K., Johannesson, B. and Rolán-Alvarez, E. 1993.
Morphological differentiation and genetic cohesiveness over
a microenvironmental gradient in the marine snail Littorina
saxatilis. / Evolution 47: 1770 /1787.
Johannesson, K., Rolán-Alvarez, E. and Ekendahl, A. 1995.
Incipient reproductive isolation between two sympatric
morphs of the intertidal snail Littorina saxatilis. / Evolution 49: 1180 /1190.
Johannesson, K., Rolán-Alvarez, E. and Erlandsson, J. 1997.
Growth rate differences between upper and lower shore
ecotypes of the marine snail Littorina saxatilis (Olivi)
(Gastropoda). / Biol. J. Linn. Soc. 61: 267 /279.
Kotler, B. P. 1984. Risk of predation and the structure of desert
rodent communities. / Ecology 65: 689 /701.
Lawlor, L. R. and Maynard Smith, J. 1976. The coevolution and
stability of competing species. / Am. Nat. 110: 79 /99.
Longland, W. S. and Price, M. V. 1991. Direct observations of
owls and heteromyid rodents: can predation risk explain
microhabitat use? / Ecology 72: 2261 /2273.
MacArthur, R. and Levins, R. 1967. The limiting similarity,
convergence, and divergence of coexisting species. / Am.
Nat. 101: 377 /385.
McCormack, S. M. D. 1982. The maintenance of shore-level
size gradients in an intertidal snail (Littorina sitkana ).
/ Oecologia 54: 177 /183.
McLaughlin, R. L., Ferguson, M. M. and Noakes, D. L. G.
1999. Adaptive peaks and alternative foraging tactics:
evidence of short-term divergent selection for sitting-andwaiting and actively searching. / Behav. Ecol. Socio. 45:
386 /395.
Murphy, P. G. 1976. Electrophoretic evidence that selection
reduces ecological overlap in marine limpets. / Nature 261:
228 /230.
Peterson, C. H. and Black, R. 1994. An experimentalist’s
challenge: when artifacts of intervention interact with
treatments. / Mar. Ecol. Prog. Ser. 111: 289 /297.
Reid, D. G. 1996. Systematics and evolution of Littorina . / Ray
Society No. 164, London.
Robles, C., Sweetnam, D. A. and Dittman, D. 1989. Diel
variation of intertidal foraging by Cancer productus L. in
British Columbia. / J. Nat. Hist. 23: 1041 /1049.
524
Rochette, R. and Dill, L. M. 2000. Mortality, behavior and the
effects of predators on the intertidal distribution of littorinid
gastropods. / J. Exp. Mar. Biol. Ecol. 253: 165 /191.
Rochette, R., Dunmall, K. and Dill, L. M. 2003. The effect of
life-history variation on the population size structure of a
rocky intertidal snail (Littorina sitkana ). / J. Sea Res. 49:
119 /132.
Rolán-Alvarez, E., Rolán, E. and Johannesson, K. 1996.
Differentiation in radular and embryonic characters,
and further comments on gene flow, between two
sympatric morphs of Littorina saxatilis (Olivi). / Ophelia
45: 1 /15.
Sacchi, C. F. 1969. Recherches sur l’écologie comparée de
Littorina obtusata (L.) et de Littorina mariae Sacchi et Rast
(Gastropoda: Prosobranchia) en Galice et en Bretagne.
/ Inv. Pesq. 33: 381 /413.
Sacchi, C. F. and Voltolina, D. 1987. Recherches sur l’écologie
comparée des littorines dans l’ı̂le de Vancouver (British
Columbia, Canada). / Atti Soc. Ital. Sci. Nat. Mus. Civ.
Stor. Nat. Milano. 128: 209 /234.
Sokolova, I. M. and Pörtner, H. O. 2001. Temperature effects on
key metabolic enzymes in Littorina saxatilis and L. obtusata
from different latitudes and shore levels. / Mar. Biol. 139:
113 /126.
Sokolova, I. M., Bock, C. and Pörtner, H. O. 2000. Resistance
to freshwater exposure in White Sea Littorina spp. I:
Anaerobic metabolism and energetics. / J. Comp. Physiol.
170: 91 /103.
Sutherland, W. J. 1996. From individual behaviour to population ecology. / Oxford Univ. Press.
Underwood, A. J. 1979. The ecology of intertidal gastropods.
/ Adv. Mar. Biol. 16: 111 /210.
Vamosi, S. M. 2002. Predation sharpens the adaptive peaks:
survival trade-offs in sympatric sticklebacks. / Ann. Zool.
Fenn. 39: 237 /248.
Vincent, T. L. S., Scheel, D., Brown, J. S. and Vincent, T. L.
1996. Trade-offs and coexistence in consumer-resource
models: it all depends on what and where you eat. / Am.
Nat. 148: 1038 /1058.
Wcislo, W. T. 1989. Behavioral environments and evolutionary
changes. / Annu. Rev. Ecol. Syst. 20: 137 /169.
Zar, J. H. 1984. Biostatistical analysis, 2nd ed. / PrenticeHall.
Zimmer-Faust, R. K., Fielder, D. R., Heck, K. L. J. et al. 1994.
Effects of tethering on predatory escape by juvenile blue
crabs. / Mar. Ecol. Prog. Ser. 111: 299 /303.
OIKOS 105:3 (2004)
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertising